Effect of the grinding behaviour of coal blends on coal utilisation for combustion
F. Rubieraa, A. Arenillasa, E. Fuentea, N. Milesb and J.J. Pisa*
a
Instituto Nacional del Carbón. CSIC. Apartado 73. 33080 Oviedo. Spain.
b
School of Chemical, Environmental and Mining Engineering. Nottingham University. NG7
2RD. United Kingdom.
Abstract
Grinding of a high volatile bituminous coal was performed in three comminution devices:
Raymond Mill, Rolls Crusher and Ball Mill. The pulverised samples were sieved to obtain
four particle size fractions and temperature-programmed combustion (TPC) was used for the
evaluation of their combustion behaviour. In addition, three coals of different hardness and
rank were mixed in various proportions in order to compare the combustibility characteristics
of the binary coal blends with those of the individual coals. The effect of coal blending on
grindability was also studied. It was found that grindability was non-additive especially when
coals of very different Hardgrove Grindability Index (HGI) were blended. The combustion
studies also suggested that there exist interaction between individual coals when they are
burnt as a blend.
Résumé
Un charbon bitumineux contenant un taux élevé de matières volatiles a été broyé au moyen
de trois types de broyeurs: un broyeur Raymond, un broyeur à rouleaux et un broyeur à
boules. Les échantillons pulvérisés ont été tamisés pour obtenir quatre fractions de différent
diamètres de particules et des essais de combustion à température contrôlée (TPC) ont été
réalisés pour évaluer leur comportement pendant le processus de combustion. En outre, trois
charbons de dureté et rang différents ont été mélangés en proportions variées, afin de
comparer les caractéristiques de combustion des mélanges binaires à celles des charbons
individuels. L’effet du mélange des charbons sur la capacité de mouture a été également
étudié. On a trouvé que cette capacité n’est pas additive, surtout quand des charbons ayant des
Indices Hardgrove (HGI) très différents sont mélangés. Les études de combustion montrent
1
aussi qu’il existe une intéraction entre les charbons individuels quand ils sont brûlés en tant
que mélange.
Keywords: Comminution; Hardgrove Index; Coal blends; Combustion
1. Introduction
Coal grindability is a complex property related to coal hardness, strength, tenacity, and
fracture. All these properties are influenced by coal rank, petrography, and the type and
distribution of minerals. The grindability of a coal blend, as measured by the Hardgrove
Grindability Index (HGI), is of great interest since it is used as a predictive tool to determine
the capacity of industrial pulverisers in coal-fired power stations. There have been some
investigations on the HGIs of coal blends specifically in relation to the additivity of the HGI
and although some coal blends show additivity this is not usually the case [1-3]. Thus, there is
no general method for predicting the HGI of a coal blend, and hence, it must be determined
experimentally on a case by case basis. This, in turn, makes it very difficult to set the
grindability specification of the coals to be blended. Utilities are, in addition, switching to
low-NOx combustion technologies and need to obtain a balance between the conflicting
requirements of minimum NOx emissions and acceptable unburnt carbon levels in the fly ash
[4]. This dependence of unburnt carbon and NOx emissions could potentially be minimised
by grinding the coal finer which in turn will be dependant upon the method of comminution.
In this respect there have been few studies carried out on the comminution of coal blends, and
the resulting products, by different methods of crushing and grinding [5]. Thus a better
understanding of the effects of the size distribution, Hardgrove Grindability Index and coal
blending on pulverised fuel behaviour would be of great benefit.
Coal blending is of increasing interest to power stations firing pulverised coal, as operators
attempt to increase the flexibility of fuel types, improve the combustion behaviour of their
coals and meet the requirements of emission legislation. Some aspects of the combustion
behaviour of blended coals in power station boilers are known, and can be determined
reasonably well from knowledge of the properties of the component coals in the blend.
However, the effects of firing blended coals on ignition behaviour, NOx emissions, burnout
and the incidence of phenomena such as slagging and fouling cannot be so easily inferred by
this approach, particularly if low NOx technologies are employed [6].
2
Additionally, the difficulties in burning low volatile coals and anthracites under NOx
control conditions are well known. The increase in volatile species in the flame zone by
blending these types of coal with high volatile coals can substantially reduce NOx emissions
favouring, at the same time, the combustion behaviour of the higher rank coal.
The main objective of this work was to study the effect of coal particle size distribution on
combustion behaviour with particular attention being given to the effect of mixing coals of
different rank and varying Hardgrove Grindability Index. To this end four coals of varying
rank from high volatile bituminous coal to anthracite were chosen as starting materials. One
of the coals was ground in several comminution devices and dry screened to obtain different
particle size fractions. A differential thermogravimetric system was utilised for the evaluation
of the combustion behaviour of the single coals and their blends.
2. Experimental
2.1. Sample preparation
Three coals of varying rank from high volatile bituminous coal to anthracite were selected
for the preparation of binary blends. Their proximate analysis is given in Table 1. LO and WI
were USA coals while GI was a Spanish anthracite. A low ash content was another criteria
used for the selection of the samples in order to minimise the effects of mineral matter.
Preparation of binary coal blends for evaluation of combustion behaviour, involved grinding
the individual coals in a mortar grinder to minus 150 µm in size and dry sieving to obtain a
75-150 µm size fraction. The individual coals were mixed after grinding by varying the
content of lower rank coal to generate three blends containing 25%, 50% and 75% by weight
respectively.
The Hardgrove Grindability Index (HGI) was measured in accordance with standard
procedures [7]. The HGIs of the binary coal blends were determined after mixing 0.60-1.18
mm size fractions in the proportions outlined above.
3
2.2. Comminution
Three comminution devices – Raymond Mill (RM), Rolls Crusher (RC) and Ball Mill
(BM) - were used to grind a UK high volatile bituminous coal (UK1). A representative
sample was split from the bulk 2 tonne sample of -50 mm material. This sub-sample was
staged crushed in a jaw crusher set at 20 mm and then to 6 mm. This material was then
crushed in a rolls crusher set at 2 mm. The above material was then split into three subsamples and comminuted in the following three devices:
Raymond Mill - This was a laboratory swing hammer mill fitted with an internal 212 µm
screen. Breakage was predominantly by impact of the hammers on the coal.
Rolls Crusher – This was a Strutevant double smooth rolls crusher, 102 mm in length and 203
mm in diameter.This crusher was set to its minimum gap of around 1 mm and the material fed
through. The resultant product was screened on a 212 μm sieve with the oversize re-crushed
at the same set. The procedure was repeated 8 times.
Ball Mill - The coal sample was dry ground in a 305x305 mm laboratory ball mill. Total
weight of steel ball charge was 31.6 kg, which occupy 50% of the mill volume and
comprising three ball sizes: 3.8 cm diameter (24.8 kg), 5.1 cm diameter (6.1 kg) and 0.9 cm
diameter (0.7 kg). The material was ground for 10 minutes and then sieved on a 212 μm sieve.
The resulting products from each device were then dry sieved to generate the following
discrete size fractions; 150-212 µm, 75-150 µm, 38-75 µm and <38 µm.
2.3. Temperature-programmed combustion tests
Temperature-programmed combustion profiles (TPC) were obtained using a differential
thermogravimetric analyser (Setaram TAG 24). It is well known that the use of different
equipment and operational conditions give significant variations in the results obtained by
TGA. Experimental conditions leading to consistent reproducible results were already
established from previous work [8, 9]. In all experimental work approximately 25 mg of coal
and an air flow rate of 50 cm3 min-1 were used. The samples were heated at 15°C min-1 from
room temperature to 1200°C. A number of parameters can be derived from TPC profiles and a
typical combustion profile is shown in Figure 1. The rate of weight loss curve (DTG), the rate
of heat flow curve (DTA) along with the main characteristic parameters that will be used in
this work are displayed in Figure 1. These parameters are defined as follows:
4
Tv
Volatile matter initiation temperature. Calculated as the temperature where the rate of
weight loss reached the value of 0.005% s-1 following the loss of moisture and oxygen
chemisorption.
Tm Temperature of maximum rate of weight loss or peak temperature.
Rmax The reactivity value at the peak temperature.
T1/2 Temperature at 50% burn-off of organic material.
Te
Temperature of char burnout. Calculated as the temperature where the rate of weight
loss reached a value of 0.005% s-1.
tb
Burnout time or time between Tv and Te.
Tsh Temperature of self-heating. Calculated from the DTA curve as the temperature where
the rate of heat flow began to be exothermic.
Tec End of combustion temperature. Calculated as the temperature where the rate of heat
flow passed from exothermic to endothermic.
3. Results and discussion
3.1. Effect of comminution on coal combustion behaviour
Temperature-programmed combustion tests were conducted on the 12 coal samples
obtained from the comminution devices. These results are presented in Figure 2 as a series of
curves showing the variation of the rate of weight loss against temperature. The most relevant
characteristic parameters from the TPC profiles are given in Table 2.
It can be seen from Figure 2 a shift of the curves to higher temperatures as the particle size
increased. The volatile initiation temperature, Tv, increased as did the particle size; this
corroborates the well-known poorer ignition and burning behaviour encountered with larger
particle sizes. Other temperatures such as Te, Tm and T1/2 were even more affected by an
increase in particle size. For instance, in the case of the Rolls Crusher the temperature Te,
passed from 621°C for the smallest fraction (-38 µm) to 729°C for the largest fraction (150212 µm). The increase in these parameters along with that of the burnout time, tb, also
indicated that the fine particles required lower residence times and temperatures to achieve
complete combustion.
5
Another feature that can be observed in Figure 2 is the stage of oxygen chemisorption,
which occurred after the loss of moisture. There was in all cases a net increase in the amount
of chemisorbed oxygen with decreasing particle size. This was a consequence of the increase
in the external surface area of the finer particles with more active sites exposed to the action
of the reactive gases.
The self-heating temperature, Tsh, is often regarded as an indicator of the susceptibility of
coal to self-heating and spontaneous ignition. It has been observed to increase with decreasing
coal rank, with lignite and subbituminous coals being more susceptible to self-heating than
bituminous coals and anthracites [8, 10]. As can be seen in Table 2 the temperatures of selfheating generally increased in the samples from the Rolls Crusher and Raymond Mill.
However the values for the material from the Ball Mill are slightly lower and fairly constant
over the various size ranges at a value of around 191°C. Nevertheless, these Tsh values were
relatively low which means that there might be some risk of spontaneous combustion if there
was a high proportion of fine material in stockpiles of this coal. The possible hazardous
effects such as fires and explosions in the milling systems are another matter of concern [11].
A relative ranking in terms of combustibility can be established from the above results.
Usually the burnout temperature, Te, and the maximum reactivity, Rmax, are taken as
indicators of combustion behaviour for comparative purposes. From Figure 2 it can be
observed that the samples obtained from the Ball Mill were more reactive and exhibited a
better combustibility than those from the Rolls Crusher which, in turn, were better than those
from the Raymond Mill. The reasons for this are unclear but may be a function of the detailed
particle size distributions of all the products and the associated quantities of mineral matter
present in the discrete size fractions.
Another feature observed in the 150-212 µm size fractions, particularly those from the
Rolls Crusher and Raymond Mill (Figure 2), was the preliminary peak that appeared before
the peak temperature. The preliminary peak is due to the combustion of volatiles, which occur
at the surface of the material, and the reaction is not sufficient to sustain the combustion of the
char, with two peaks appearing in the combustion profiles. The presence of this peak in the
150-212 µm fraction may be explained as a consequence of the segregation of macerals
among the size fractions, with a tendency for the liptinite macerals, which yield the greatest
amount of volatiles, to concentrate to the coarser size fractions [12]. Similarly, the
6
agglomeration of particles due to the development of thermoplasticity has also been suggested
as a cause of the preliminary volatiles evolution peak [13]. It was then decided to carry out
additional combustion tests on the samples that exhibited the volatiles combustion peak.
These 150-212 µm size fractions were re-ground to minus 75 µm and the results for the
material from the Rolls Crusher are illustrated in Figure 3. It can be seen that the combustion
profiles of the re-ground fraction was different to that of the original one, with the
disappearance of the preliminary volatiles combustion peak. The profile for the 38-75 µm
fraction from the Rolls Crusher has been also included in Figure 3 for comparative purposes.
It can be observed that the resultant combustion profile of the re-ground fraction clearly
resembled that for the 38-75 µm sample and not the original 150-212 µm fraction. It can be
concluded that although swelling of the larger particles may have taken place, the volatile
evolution peak was a function of the coarser particle sizes.
3.2. Grindability and combustion behaviour of coal blends
The effect of mixing coals of different hardness and rank on the HGIs and combustion
behaviour of the blends, in comparison with those of the individual coals, was also studied.
Three different rank coals (see Table 1) were used for these tests. HGI data of the single coals
and the blends are shown in Figure 4. Coals GI and LO were hard coals according to their
HGIs of 34 and 47 respectively, whilst coal WI was a soft coal with a HGI of 91. From the
results represented in Figure 4 it can be seen that the HGIs are highly non-additive when
considering the blends of soft and hard coals (WI/LO and WI/GI). This is in disagreement
with the results found by others using the same type of blends [3]. The blend made up of the
two hard coals (GI/LO) showed a more linear relationship. Nevertheless, in all cases the HGI
of the blend could not be predicted from the individual HGI of a coal and the weight fractions
used in the blends. It is apparent that there was a preferential shift of the HGI of the blend
towards higher values than would be expected from the linear additive rule. This indicates
that when grinding blends comprising the individual coals of WI and LO or WI and GI there
will be a segregation of the LO and GI coals to the smaller size fractions. From a practical
point of view this could have important implications in the combustion behaviour of blends
made from these coals. For instance the blending of the coal LO with the anthracite GI could
be desirable in order to favour the combustibility of the higher rank coal, GI, and to reduce
NOx emissions by increasing the amount of volatile species in the flame zone. The fact that
7
grindability is non-additivite could produce a blend from the industrial pulveriser that is out of
specification to that previously established.
The combustion behaviour of the three coals and their blends was evaluated by
temperature-programmed combustion tests. Binary coal blends were prepared using the 75150 µm size fractions of the individual coals. The use of a relatively large particle size
fraction, 75-150 µm, is of significance for predicting combustion behaviour, especially carbon
burnout. Most current analysis techniques assess bulk properties, but for burnout prediction
the behaviour of only a small fraction of particles, the ‘least-likely-to-burn’ or LLB fraction,
is considered more relevant [14].
The results of the combustion tests for the LO/GI coal blends are shown in Figure 5. It can
be seen that the volatile initiation temperature increased as the anthracite (GI) content in the
blend increased. The maximum temperature remained closer to the peak temperature value of
the coal with a higher percentage in the blend. However, at 50% by weight of anthracite in the
blend it was possible to distinguish, in the zone of char combustion, the burning of the coal,
LO, from that of the anthracite. This behaviour differs from that found by others who reported
that the burning profile of blends of coals of different rank showed two well resolved peaks
[15]. In addition, the burnout temperature, Te, increased as the percentage of anthracite in the
blend increased but its value was always lower than that of the anthracite. Similar results were
obtained for the blends, WI/GI and LO/WI. This indicated that some interaction between the
samples has occurred during temperature-programmed combustion tests.
4. Conclusions
Evaluation of the combustion behaviour of a high volatile bituminous coal, comminuted in
three different devices, has indicated that the samples pulverised in the Ball Mill exhibited
better combustibility characteristics than those obtained from the Raymond Mill and the Rolls
Crusher. The reason for this is unclear but may be a function of the particle size distributions.
The HGI of binary coal blends could not be predicted from the weighted average of the
HGIs of the individual coals in the blend. This was especially true in the case of mixing a soft
anthracite with a hard coal. The HGI of the blend resembled that of the softer coal. This may
have important implications in combustion performance as the feed to a pulverised-fuel power
8
station using these type of blends could be quite different than specified. One potential
method of avoiding this problem would be to blend the coals after pulverisation.
Temperature-programmed combustion tests carried out in this work with three binary coal
blends suggested that the coals burn with interaction. The combustion profile and the
characteristic parameters of the blends could not be predicted from the data of the individual
coals.
Acknowledgements
Work carried out with a financial grant from the European Coal and Steel Community
(Project 7220-EA/133).
References
[1]
J.C. Hower, J. of Coal Quality, 7 (1988) 68.
[2]
R.E. Douglas, M.C. Mai and R.D. Stoessner, Proceedings 12th Int. Pittsburg Coal
Conference, (1990) pp. 837-851.
[3]
A. Waters, J. of Coal Quality, 5 (1986) 33.
[4]
H. Maier, H. Spliethoff, A. Kicherer, A. Fingerle and K.R. Hein, Fuel, 73 (1994) 1447.
[5]
S.F. Scieszka, Powder Technol., 49 (1987) 191.
[6]
J.P. Smart, J.P and T. Nakamura, J. Institute of Energy, 6 (1993) 99.
[7]
ASTM D 409, Standard test method for grindability of coal by the Hardgrove-Machine
method.
[8]
J.J. Pis, G. de la Puente, E. Fuente, A. Morán and F. Rubiera, Thermochimica Acta, 279
(1996) 93.
[9]
F. Rubiera, A. Morán, O. Martínez, E. Fuente and J.J. Pis, Fuel Proc. Technol., 52
(1997) 165.
9
[10] J.M. Kuchta, V.R. Rowe and D.S. Burgess, Report Invest. U.S. Bureau of Mines, No.
8474 (1980).
[11] D.H. Scott, Coal pulverisers-performance and safety, IEA Coal Research, IEACR/79,
London (1995).
[12] A. Davis and P.C. Orban, Proceedings 12th International Pittsburg Coal Conference,
(1995) pp. 312-317.
[13] J.B. Milligan, K.M. Thomas and J.C. Crelling, Fuel, 76 (1997) 1249.
[14] J.R. Gibbins and J. Williamson, Proceedings 7th International Conference on Coal
Science, Banff, Canada, (1995) Vol. 1, pp.35-38.
[15] J.W. Cumming, Thermochimica Acta, 155 (1989) 151.
10
Table 1. Main characteristics of the coals used for the preparation of binary coal blends.
Coal
LO
WI
GI
Moisture (wt %)
1.4
0.7
3.0
Ash (wt %, db)
6.7
8.0
9.3
V.M. (wt %, db)
34.1
17.4
3.8
F.C. (wt %, db)
59.2
74.6
86.9
C.V. (kcal/kg, db)
7839
8037
7396
Hardgrove Index
47
91
34
Proximate Analysis
(db: dry basis)
11
Table 2. Characteristic parameters from temperature-programmed combustion tests for the
different size fractions obtained from the three comminution devices.
Rmaxx102 Tsh
Size
Comminution
Tv
Tm
Te
T1/2
tb
Fraction
Device
(°C)
(°C)
(°C)
(°C)
(s)
(%/s)
(°C)
(°C)
RC
325
532
729
556
1595
8.95
221
730
RM
325
544
745
561
1657
8.36
214
748
BM
324
544
732
559
1606
8.78
194
738
RC
322
523
717
540
1561
8.71
194
722
RM
320
523
741
552
1658
8.60
216
743
BM
315
514
685
529
1457
9.89
192
689
RC
313
519
674
524
1424
10.30
195
677
RM
316
519
706
519
1537
9.33
206
707
BM
302
495
640
496
1333
10.88
188
646
RC
299
485
621
484
1268
11.87
186
630
RM
301
485
622
484
1267
11.94
193
624
BM
298
469
600
468
1192
12.11
191
604
Tec
(µm)
150-212
75-150
38-75
-38
12
List of Figures
Figure 1. Typical combustion profile showing characteristic parameters.
Figure 2. Comparison between the combustion profiles of the different size fractions obtained
in the crusher and the two mills.
Figure 3. Combustion profiles of the 38-75 µm, 150-212 µm fractions from the Rolls Crusher
and the re-ground material (-75 µm).
Figure 4. Hardgrove grindability indexes of the individual coals and their blends.
Figure 5. Combustion profiles of blends of the coal LO and the anthracite GI.
13
Figure 1. Typical combustion profile showing characteristic parameters.
14
Figure 2. Comparison between the combustion profiles of the different size fractions obtained
in the crusher and the two mills.
15
Figure 3. Combustion profiles of the 38-75 µm, 150-212 µm fractions from the Rolls Crusher
and the re-ground material (-75 µm).
16
Figure 4. Hardgrove grindability indexes of the individual coals and their blends.
17
Figure 5. Combustion profiles of blends of the coal LO and the anthracite GI.
18